In tangchao7498/DecodeR: Barcode demultiplexing of native RNAs nanopore sequencing

knitr::opts_chunk$set(echo = TRUE, warning = FALSE, message = FALSE, fig.align = 'center')

library(data.table)
library(caret)

Reads of Poreplex & DeePlexiCon barcodes

read2gene_20200605 <- fread("/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex/analysis/02.BamReadsSplit/read2gene_20200605.txt")
read2gene_20200620 <- fread("/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex/analysis/02.BamReadsSplit/read2gene_20200620.txt")
read2gene_20200902 <- fread("/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex/analysis/02.BamReadsSplit/read2gene_20200902.txt")
read2gene_20201127 <- fread("/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex/analysis/02.BamReadsSplit/read2gene_20201127.txt")
read2gene_20210703 <- fread("/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex/analysis/02.BamReadsSplit/read2gene_20210703.txt")

read2gene_20200605[, read := paste0("read_", read)]
read2gene_20200620[, read := paste0("read_", read)]
read2gene_20200902[, read := paste0("read_", read)]
read2gene_20201127[, read := paste0("read_", read)]
read2gene_20210703[, read := paste0("read_", read)]

read2gene_20200605[, table(Name)]
read2gene_20200620[, table(Name)]
read2gene_20200902[, table(Name)]
read2gene_20201127[, table(Name)]
read2gene_20210703[, table(Name)]

read2gene_20200902 <- read2gene_20200902[grepl("^[DP]-", Name), ]
read2gene_20201127 <- read2gene_20201127[grepl("^[DP]-", Name), ]
read2gene_20210703 <- read2gene_20210703[grepl("^[DP]-", Name), ]

read2gene <- rbind(data.table(Batch = "20200605", read2gene_20200605), 
                   data.table(Batch = "20200620", read2gene_20200620), 
                   data.table(Batch = "20200902", read2gene_20200902), 
                   data.table(Batch = "20201127", read2gene_20201127), 
                   data.table(Batch = "20210703", read2gene_20210703))

dir_bs <- "/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex/analysis/03.BarcodeProcessing/02.BarcodeSignalBinExp/"
files <- list.files(path = dir_bs, pattern = ".signal", full.names = TRUE, recursive = TRUE)

Sigs <- mclapply(FUN = function(x) {
  load(x)
  barcode
}, files, mc.cores = 20)
Sigs <- do.call("c", Sigs)

Sigs <- Sigs[names(Sigs) %in% read2gene[, read]]
read2gene <- read2gene[read %in% names(Sigs)]

read2gene[, table(Name)]

Training and test set of DeePlexiCon & Poreplex

DeePlexiCon

For DeePlexiCon software

DeePlexiCon_TrainingSet <- read2gene[grepl("^D-", Name), .SD[sample(.N, 10000)], by = "Name"]
DeePlexiCon_TestSet <- read2gene[grepl("^D-", Name) & !read %in% DeePlexiCon_TrainingSet[, read], ]

Poreplex

For Poreplex software

Poreplex_TrainingSet <- read2gene[grepl("^P-", Name), .SD[sample(.N, 100000)], by = "Name"]
Poreplex_TestSet <- read2gene[grepl("^P-", Name) & !read %in% Poreplex_TrainingSet[, read], ]

Signals of training and test set of DeePlexiCon & Poreplex

DeePlexiCon

DeePlexiCon_TrainingSet_Signal <- Sigs[DeePlexiCon_TrainingSet[, read]]
DeePlexiCon_TrainingSet_Signal <- as.data.frame(do.call(rbind, DeePlexiCon_TrainingSet_Signal))
colnames(DeePlexiCon_TrainingSet_Signal) <- paste0("BIN", sprintf("%03d", seq_len(ncol(DeePlexiCon_TrainingSet_Signal))))
DeePlexiCon_TrainingSet_Signal$Class <- DeePlexiCon_TrainingSet[, as.factor(Name)]

DeePlexiCon_TestSet_Signal <- Sigs[DeePlexiCon_TestSet[, read]]
DeePlexiCon_TestSet_Signal <- as.data.frame(do.call(rbind, DeePlexiCon_TestSet_Signal))
colnames(DeePlexiCon_TestSet_Signal) <- paste0("BIN", sprintf("%03d", seq_len(ncol(DeePlexiCon_TestSet_Signal))))
DeePlexiCon_TestSet_Signal$Class <- DeePlexiCon_TestSet[, as.factor(Name)]

Poreplex

Poreplex_TrainingSet_Signal <- Sigs[Poreplex_TrainingSet[, read]]
Poreplex_TrainingSet_Signal <- as.data.frame(do.call(rbind, Poreplex_TrainingSet_Signal))
colnames(Poreplex_TrainingSet_Signal) <- paste0("BIN", sprintf("%03d", seq_len(ncol(Poreplex_TrainingSet_Signal))))
Poreplex_TrainingSet_Signal$Class <- Poreplex_TrainingSet[, as.factor(Name)]

Poreplex_TestSet_Signal <- Sigs[Poreplex_TestSet[, read]]
Poreplex_TestSet_Signal <- as.data.frame(do.call(rbind, Poreplex_TestSet_Signal))
colnames(Poreplex_TestSet_Signal) <- paste0("BIN", sprintf("%03d", seq_len(ncol(Poreplex_TestSet_Signal))))
Poreplex_TestSet_Signal$Class <- Poreplex_TestSet[, as.factor(Name)]

setwd("/mnt/raid5/Personal/tangchao/project/Nanopore/BarcodeDecomplex/analysis/06.DecodingDeePlexiConAndPoreplex/00.ModelTraining2/")
saveRDS(DeePlexiCon_TrainingSet_Signal, file = "00.RData/DeePlexiCon_TrainingSet.Rds")
saveRDS(DeePlexiCon_TestSet_Signal, file = "00.RData/DeePlexiCon_TestSet.Rds")

saveRDS(Poreplex_TrainingSet_Signal, file = "00.RData/Poreplex_TrainingSet.Rds")
saveRDS(Poreplex_TestSet_Signal, file = "00.RData/Poreplex_TestSet.Rds")

saveRDS(read2gene, file = "00.RData/read2gene.Rds")